of Cyclically Varying Temperature, 289 



strength that, if an abnormal explosion occurred, it was in- 

 variably destroyed. 



Calibration of the Thermometers. 



The ice points call for no special remark beyond the fact 

 that, in spite of considerable care in ensuring a mixture of 

 finely crushed ice and water, they were by no means as satis- 

 factory as the steam points. The average resistance of the 

 0-003, 0-0025, and 0*002 inch diameter wires was 0'3, 0*55, 

 and 0*75 ohm respectively. The steam points were obtained 

 in a double-jacketed hypsometer of the ordinary form, and 

 were corrected for barometric pressure when necessary. 

 When working with these very fine wires, it is imperative to 

 guard against the heating effect of the battery current, other- 

 wise discrepant and inaccurate results are obtained. I have 

 found that the Daniell cell, mentioned above, with resistance 

 in the external circuit gave no appreciable heating unless the 

 observations were made too rapidly. (This of course only 

 applies to the ice and steam points ; when in the engine battery 

 heating is of no importance.) 



Two methods were used to obtain the quantity 8. The first 

 method was a comparison of the fine-wire thermometer with 

 a standard Callendar platinum thermometer, both being 

 placed in a tube heated to a high temperature, thus getting a 

 third point on the curve connecting the air and platinum 

 scales. 



The standard thermometer consisted of a spiral of platinum 

 wire of a diameter of 0*008 inch wound on a flat plate of mica 

 and enclosed in a porcelain tube 50 centimetres in length, 

 the flat spiral occupying a length of about 7 centimetres from 

 the closed end of the platinum tube. The fixed points of this 

 thermometer were determined manv times in ice and steam, 

 and several times in sulphur and aniline vapours. The 

 values were 



R =2'936 ohms, 

 Ri = 4*080 ohms, 

 8=1-52. 



I have not been able to detect any change in the zero, 

 although this thermometer has been raised to a bright red 

 heat several hundred times during the last two years. 



For the comparison of the two thermometers an ordinary 

 gas-furnace, such as is used for combustion work, was first 

 used, but was abandoned owing to unequal and unsteady 

 heating. After some trials I adopted a tube-furnace heated 

 with gas-coke ; from end to end of the furnace (about 2 feet 

 in length) was placed a cast-iron tube 2-| inches diameter 



